Shitagaki Lab see-through Turns Anime Art Into Animatable Layers
see-through automatically converts flat anime drawings into editable, multi-layered files suitable for animation. The system splits a single picture into up to twenty-three separate parts, like hair, eyes, and clothing, while intelligently filling in hidden background areas.
Developed by researchers at Shitagaki Lab and Saint Francis University, this tool tackles the tedious task of manual image segmentation. Small studios can use it to prepare static illustrations for dynamic projects without starting from scratch.
Model Size: not specified & VRAM GPU: 8-16GB required
Automated image layering system
- Splits input drawings into up to twenty-three distinct semantic parts.
- Fills missing background sections using transparent layer generation.
- Generates pseudo-depth maps tailored for animated artwork.
- Exports organized Photoshop files ready for immediate editing.
- Supports operation on lower-end hardware through specialized memory management.
Creators handling large batches of character art will save hours normally spent on manual masking and background reconstruction. The automated workflow allows teams to quickly move from flat concepts to rigged assets without needing advanced programming skills or specialized editing software.
Current capabilities and technical limits
Processing head and body layers runs in separate phases, which extends the overall computation time. The team emphasizes that the current output serves as a structural foundation rather than a complete animation package.
"Automating one step in isolation cannot replicate this,"
the developers explained regarding complex workflow requirements in a project readme. Future updates will focus on expanding language options and releasing training documentation by April 2026.
The project provides a reliable method for breaking down complex illustrations while maintaining original artistic details. You can access the source code and setup guides on GitHub.
